scenario-sketchpad

Navigraph is an advanced visual orchestration platform for designing, simulating, and validating complex operational scenarios for autonomous vehicles, robotic fleets, and intelligent systems. It transforms scenario planning into a dynamic, interactive canvas where logic meets visualization, serving engineers, researchers, and system architects. Key capabilities include dynamic visual scripting with a node-based interface, multi-agent orchestration for defining behaviors of numerous independent agents, parametric environment modeling with adjustable time, weather, and road conditions, a real-time validation engine that checks for inconsistencies, and export to simulation-ready formats like OpenSCENARIO, ROS 2 world files, and CARLA JSON. It also supports collaborative canvas editing with real-time cursors and version history, and a responsive UI that scales from multi-monitor workstations to tablets. Getting started requires Python 3.10+, Node.js 18+, and a modern browser. Installation is available via direct download, pip (`pip install navigraph`), or Docker. The CLI allows launching a local server, initializing projects from templates, and exporting scenarios to formats like OpenSCENARIO. Navigraph integrates with AI platforms like OpenAI and Claude for generating natural language descriptions, creating agent dialogue trees, and reviewing scenarios for safety edge cases. These integrations are opt-in and respect configured privacy levels. The platform includes an infinite layer-based canvas, template and snippet libraries, Git-like version control, performance metrics overlay during preview simulation, multilingual UI support, community support, and a plugin ecosystem via an SDK. Navigraph is provided "as is" and intended for simulation, design, and testing in controlled environments. Scenarios must be reviewed by qualified engineers before deployment in safety-critical systems. It is licensed under the MIT License.